Frequently Asked Questions

Finches is the intelligence layer for agricultural sourcing.

We connect the data that actually decides your raw material: weather, satellite, agronomic science, soil, market and trade rules, and local news, with your own contracts, suppliers, and origins, and turn it into a sourced, graded answer you can act on.

The batch that fails your spec is usually decided weeks before it reaches you, in data you could already see. Finches connects that data into a sourcing decision while you can still act, for any commodity, from your desk.

The knowledge to prevent most sourcing losses already exists. It is just not connected.

Around €184bn is lost across agricultural supply chains every year to drought, pests, tariffs, and quality failures, and in most cases the warning signs were sitting in separate systems the whole time: a weather model knew the rain was coming, agronomy knew the flowering window, satellite saw the standing water, a trade rule knew a cheaper origin was duty-free. None of it resolved against your contract in time.

Finches closes that gap, so a forming risk reaches your sourcing team as a specific answer instead of a surprise at harvest.

• No early visibility into crop, climate, and supplier risk

• Slow, manual agronomy and procurement workflows

• Little transparency from field to factory

• Reactive firefighting instead of planning ahead

Item descriptionFinches is built on a connected model of your supply base, not a stack of separate feeds.

It holds your real-world objects: origin, region, crop, supplier, contract, and wires them to layered external data.

Because everything is linked, the system knows which supplier, batch, and contract a given event actually hits, and reasons across all of it at once instead of reacting to a single signal. You point Finches at your commodities and origins, ask a question in plain language, and get an answer that is sourced and ready to act on. The value is in the connections, not in any one single data source.

Finches gives sourcing leads early, specific answers tied to their own contracts. You can ask across your whole supply base in plain language (for example, "which of my durum origins are trending over my DON limit?"), get supplier-level risk alerts mapped to your variety specs rather than a generic regional feed, and see forming risks from weather, science, regulation, trade, and pests before they reach the lab or the invoice. The output is a sourcing decision you can act on: where to contract now, where to hold, where to re-allocate volume. Drafted actions and re-allocation proposals are the next layer we are building.

Finches gives agronomy and field teams a faster way to capture what they see and lose nothing to a notebook or a buried WhatsApp message.

With Finches Field, the mobile app, they record voice notes, photos, and video in seconds, even offline, and each visit is automatically structured into a searchable, audit-ready report with no forms to fill. Visits are prioritized by risk, so time goes to the farms that need it. And every observation becomes part of the connected model: the curl of a stressed leaf or the first hidden cluster of pest eggs turns into an early signal that sharpens the risk picture for the whole sourcing team. Field work stops disappearing into inboxes and starts compounding into intelligence.

Finches connects the outside world with your inside world.

From outside: weather forecasts and history, optical and radar satellite (plant health, soil moisture, standing water), agronomic and scientific research, pest and disease monitoring, soil data, market and price signals, trade and tariff rules, and hyperlocal news.

From inside your organization: contracts and variety specifications, lab results, ERP and sourcing history, and harvest quantity and quality.

Unstructured inputs, including scientific literature and news, are first turned into structured data the system can reason over, which is what lets a published research finding or a local outbreak report actually change an answer.

Most sourcing risk tools stop at the region. Finches stops at the supplier, at the variety, inside the contract.

Two things make that possible. First, an agricultural model that understands the domain: it knows a Brix reading on grapes is not the same as on tomatoes, that a frost in week 12 means something different for spinach than for almonds, and that durum accumulates mycotoxins differently than bread wheat. Second, it reasons across every signal at once and ties the result to your specific contracts, so you get a decision rather than another feed to watch.

A general chatbot cannot reproduce this, because the domain knowledge and the connections are the hard part.

Every answer carries its confidence. Finches grades what it tells you as verified, directional, or needs ground truth, and shows the sources behind each finding. A signed contract term or a hard lab result is verified. A forward risk built from weather and agronomic patterns is directional. Something that needs a person or a test to confirm is flagged as such. You always know what you can lean on and what still needs a look, which is what makes the intelligence usable for real sourcing decisions.

Item descriptionNo. Finches works as pure desk intelligence, and many of our examples use no field input at all. It reasons from external signals and your existing internal data, so a team sourcing pine oil or coconut from the other side of the world gets the same connected answer as one with agronomists on the ground. When you do have eyes in the field, Finches Field is an optional mobile app for field teams (agronomists, scouts, QA) to capture photos, voice notes, and video in seconds, even offline, with each visit structured into a searchable report. That ground truth sharpens a directional call into a confirmed one. It is one useful data source, not the way in.

Finches is built for companies that depend on agricultural raw materials at scale: food and beverage, baby food, textiles, coatings, and other processors exposed to climate, quality, and availability risk. We are the strongest fit where one or a few agricultural inputs are critical enough that a single quality or supply incident is felt directly on the P&L, and where procurement works to protect that supply. That is often mid-sized, family-owned processors, though the same pressures apply to the largest buyers.

Finches Intelligence is available now through our Early Access Program, where we build it together with a capped cohort of enterprise partners. It is the desktop intelligence layer: ask your data in plain language, connected external signals, and supplier-level risk alerts. Finches Field, the mobile app that feeds ground truth into it, is live and in production. And we are already going further: an operative layer that turns intelligence into drafted actions, such as volume re-allocation, is what we are building next.

Yes, but the value is not the model. Finches uses AI to do something specific: turn scattered, unstructured signals into structured data, then reason across all of it to find the patterns a human team could never catch in time. What compounds is the agricultural domain model and the connections between your objects and the outside world, refined with every customer.

A general AI assistant cannot reproduce it, because the domain knowledge and the linked model are the hard part.

Yes. Finches is built to sit on top of the systems you already run, not to replace them. We bring in your ERP and sourcing data, contracts, and lab and quality systems, and scope the specific connectors with you during onboarding, so the intelligence reflects your real supply base from the start.

Your data stays yours.

It is hosted on enterprise-grade, EU-based infrastructure and handled in line with GDPR, with SSO and role-based access controlling who sees what.

We are glad to walk your security and IT teams through the setup and to sign the standard data-processing agreements as part of onboarding.

Quickly. The first connected answers come early, as soon as your priority commodity and its signals are wired in, and they sharpen as more of your history and live data is connected. Early value shows up as risks seen weeks earlier and questions answered in minutes that used to take a team days.

Finches combines a platform fee for the intelligence layer with a per-user fee for the Finches Field mobile app. We scope the commercial model to your use case and volume in the first conversations, and partners who join the Early Access Program do so on preferential founding terms. The fastest way to a concrete number is a short call.

By catching risk earlier, Finches reduces the emergency sourcing and firefighting that quietly drive up cost and carbon. Earlier signals mean less volatility exposure, more raw material that meets quality standards, stronger and more transparent supplier relationships, and sourcing decisions you can defend on ESG.

The best first step is a short demo. We take one of your real commodities and origins, ask it a live sourcing question, and show you the answer Finches returns. From there we scope an Early Access engagement around your priority use case.

Darwin's finches (yes, the birds) showed that survival depends on detecting change early and adapting faster than everyone else. We apply the same principle to sourcing: spot the early signals in climate, crops, and supply networks, and turn them into action before shortages, quality issues, or price shocks land.

Finches was founded by Catharina van Delden (CEO), a repeat enterprise-software founder who also runs a pecan nut farm and lived through the volatility Finches now solves for; Dr. Stefanie Glenn (CTO), an AI and data expert who built systems at Palantir and Google; and Alex Vázquez Bea (CFO/COO), former CFO/COO of Oetker Digital, who led the IPO of Veganz.

The team pairs deep AI expertise with agricultural, financial, and procurement experience.

Finches began with a drought. In 2021 and 2022, Catharina watched severe drought hit her family farm, La Baguala, in Uruguay. Yields dropped, quality turned unpredictable, and decisions had to be made with far too little information, while the buyers downstream still expected certainty. The real problem was not the farming. It was the missing layer connecting what the field already knew to the sourcing decision. Finches was built to close that gap.